Starting from Emojis translation, we implement a CycleGAN with a view to translating the MNIST to USPS dataset.
This repository hosts the project for the Deep Learning course at ENSAE, made by Ban VAN TRUONG & Bertrand VUILLEMOT. It contains :
- The report summarizing two papers (Deep Convolution GANs and Wasserstein GANs) and describing the MNIST-USPS translation using CycleGANs : CycleGAN_report.pdf
- The notebook containing the code concerning the implementation of the model : cycle_gan.ipynb